Assessing the Monte Carlo Independent Column Approximation with a Super-Parametrized GCM
Barker, H.W.(a), Cole, J.N.S.(b), Khairoutdinov, M.F.(c), and Randall, D.A.(c), Meteorological Service of Canada (a), The Pennsylvania State University (b), Colorado State University (c)
Fourteenth Atmospheric Radiation Measurement (ARM) Science Team Meeting
The Monte Carlo Independent Column Approximation (McICA) method for computing radiative transfer fluxes in GCMs yields unbiased estimates of heating rate profiles with respect to the full ICA. It does so by extricating descriptions of unresolved optical structure (clouds, aerosols, surfaces...) from the radiative transfer solver. Because McICA is essentially a sampling strategy, it has as a by-product conditional random noise. Although the impact of McICA's noise on the numerical simulation of climate has been tested using a conventional GCM (see poster by Räisänen and Barker), several assumptions had to be made regarding the nature of unresolved cloud structure. This poster documents the impact of McICA's noise on a climate simulation using a GCM (NCAR-CAM version 1.8) that employed 2D cloud system-resolving models (CSRM) in each of its 8192 columns. Such a model has been referred to as a 'super-parametrized' GCM, but is known now in official circles as a 'multi-scale modelling framework' GCM, or MMF-GCM for short. One of the attractions of the MMF-GCM is that cloud structure, in particular horizontal variability and overlap, are provided explicitly down to the CSRMs' grid-spacing of 4 km. The MMF-GCM was run using both the full ICA and McICA. In both simulations, the CSRMs were provided with domain-average radiative heating profiles every 15 minutes; GCM timestep was 1 hr. As the MMF-GCM is extremely expensive to run, only 1 year simulations were conducted using specified SSTs. The simulations are still running, but 'should' be finished by the ARM-STM.
Note: This is the poster abstract presented at the meeting; an extended version was not provided by the author(s).
